|Número de publicación||US7814139 B2|
|Tipo de publicación||Concesión|
|Número de solicitud||US 11/923,608|
|Fecha de publicación||12 Oct 2010|
|Fecha de presentación||24 Oct 2007|
|Fecha de prioridad||7 Mar 2002|
|También publicado como||US7730120, US7797371, US7895258, US8055709, US8095589, US8095621, US8099496, US8626834, US8769080, US9123056, US9292860, US9501781, US9652433, US20030171977, US20080177779, US20080183717, US20080183718, US20080183796, US20080183867, US20080183868, US20080183869, US20080183870, US20110015982, US20140108096, US20150317671, US20160196567, US20170068985, US20170228771, WO2003077072A2, WO2003077072A3|
|Número de publicación||11923608, 923608, US 7814139 B2, US 7814139B2, US-B2-7814139, US7814139 B2, US7814139B2|
|Inventores||Man Jit Singh, David Cancel|
|Cesionario original||Complete, Inc.|
|Exportar cita||BiBTeX, EndNote, RefMan|
|Citas de patentes (96), Otras citas (25), Citada por (16), Clasificaciones (38), Eventos legales (6)|
|Enlaces externos: USPTO, Cesión de USPTO, Espacenet|
This application is based on, claims priority to, and incorporates by reference the following United States Provisional Patent Applications: an application filed Mar. 7, 2002 entitled Business Method for URL Analysis, naming Man Jit Singh as inventor, and an application filed Mar. 7, 2002 entitled Technology Platform for URL Analysis.
The advent of the Internet and the Worldwide Web has produced a host of electronic commerce applications, in which users interact with content and engage in a wide variety of transactions, ranging from ordering books, CDs and other items, to participating in auctions, to downloading music, to a host of other activities. Methods and systems are widely used for tracking the behavior of online users, both individually and as groups. The output from those methods and systems are typically used to adjust the structure and content of online offerings to help attract more users, or to get current users to engage in more interaction and more transactions with the provider.
Methods and systems for analyzing online user behavior range from statistical techniques, such as collaborative filtering, to use of neural nets and similar facilities. While such methods have had some success, the promise of electronic commerce remains somewhat unfulfilled. Many online businesses have failed, and those remaining find the environment increasingly competitive. Meanwhile, many businesses find that online offerings (including their own), merely take business away from offline product offerings, or that the online offerings harm offline offerings in other ways, such as by forcing price reductions. Thus, methods and systems are needed for providing improved coordination between online and offline offerings. In addition, methods and systems are needed for allowing offline businesses to take advantage of information that can be discerned from online customer behavior.
Methods and systems are provided herein for the analysis of behavior of online users. The analysis methods and systems allow for the creation of new offline business methods based on online consumer behavior.
Methods and systems are also provided for providing a host computer for handling a plurality of processes, the host system having a collection facility, a network, a pre-calculation facility, a calculation facility, and a post-calculation facility; obtaining an input data set of online user data from a plurality of data providers, wherein the input data set represents HTTP transactions of a plurality of online users; and sending an output data set to a co-location facility to permit remote access by a party other than the host.
The calculation facility may comprise providing a cluster of machines distributed in a parallel processing configuration. Output data may be stored in a data warehouse or sent to a collocation facility.
Methods and systems are also provided for obtaining an input data set from a plurality of data providers, the data set comprising data representing HTTP transactions for a plurality of users; performing a pre-calculation process to produce a file in a file format suitable for performing calculations; performing calculations on the formatted file; and performing post-calculation on the results of the calculations to produce an output file.
Methods and systems are further provided for transferring the output file to at least one of a co-location facility and a data warehouse.
Methods and systems are also provided for collecting the data set for analysis by a host computer; delivering the data set to an internal network; and performing an integrity check on the data.
Methods and systems are also provided wherein the formatted file comprises data selected from the group consisting of a file identifier, a user identifier for a user who completed an HTTP transaction, a data provider identifier for the data provider who provided the data representing the HTTP transaction, a timestamp for the HTTP transaction, a URL for the HTTP transaction, and a protocol identifier for the protocol of the transaction.
The pre-calculation step may include reading a list of URL rules; for each character of the URL of a formatted file, walking the character of the URL through the URL rules; for each rule in the URL rules, applying the rule to the URL of the formatted file; and outputting a modified URL, the modified URL reflecting the application of the rules to the original URL. The modified URL may be appended to a modified formatted file.
The methods and systems may further comprise providing a panel selection process for selecting a panel of users, data for which will be used in an input data set. The process may include conducting a survey of user demographics for an initial user panel; determining a statistic related to the demographics of the users in the user panel; comparing the statistics to the statistics for a past user panel; selecting a new panel of users representing a desired set of demographic characteristics; and determining whether to keep or discard a data item based on membership of a user in the panel.
The URL of the output file may include a session identifier, a domain identifier, a subdomain identifier, a path and a query string.
The output file may further include a hash, such as hash of the domain identifier, the subdomain identifier, or the path.
The methods and systems described herein may include calculating a metric based on the data in the output file. The metric may be a vertical market metric calculation, a search term metric calculation, or other metric.
The metric calculation may be a vertical market metric calculation that can include using a page extractor module for extracting data identifying a page of online content; using a fetcher module to retrieve the page content from a computer network; and using a classifier module to classify the page in at least one content category selected from a group of possible content categories.
Metrics may be calculated using a cluster of machines configured in a parallel-processing configuration or batch-processing configuration.
In embodiments, an input URL is decomposed based on semantic meaning of the pages viewed by the user who executes the HTTP transaction. The decomposition process may be made in the presence of redirection from a primary URL to a secondary URL or in the presence of a proxy URL for a second URL.
Methods and systems may also include obtaining an input data set reflecting clickstream activity of a plurality of online users; analyzing user behavior reflected by the clickstream activity; drawing an inference based on the user behavior; and modifying an offline process based on the inference.
The inference may relate to cross-shopping behavior between different goods and the offline process may be a marketing promotion. The input data may relate to price sensitivity of a shopper and the offline process may be a pricing process for an offline item. The input data may relate to ordering of a user's views of items and the offline process may relate to merchandizing and to product placement to promote viewing goods in a particular order. The input date may relate to user buying interest or share of interest and the offline process may relate to forecasting offline demand for a product. The input data may relate to actions of users with respect to online content choices and the offline process may relate to presenting offline media content.
In embodiments the methods and systems may include obtaining a set of data reflecting online actions of a plurality of users; analyzing the data to identify characteristics of the user's behavior; and modifying an off-line business process based on the identified characteristics.
Referring still to
The user 102 connects to online content sources by a computer network 104, such as the Internet. The online content sources typically comprise one or more servers, which may be HTTP servers. In HTTP transactions, the servers serve content over the network in response to messages from the user's computer.
In their online behavior, users can interact with sites of one or more vendors through the data provider 108. Those data providers 108 typically track users' interactions with their sites through logs or similar facilities. The logs record user behavior, tracking what pages of content they have viewed in what sequence and for what amounts of time. Thus, either at the desktop or at the server farm, information is being tracked about every HTTP transaction that the individual executes at any time. When information is tracked at a server farm it is usually subject to an agreement between the user and a host (such as an ISP or other site host). The agreement typically sets out the conditions under which the host can share data with third parties, including demographic data about the user and data about the user's actions. A facility is then put in place at the data provider 108 to follow rules that allow capturing of users' transactions and sending some individuals data while not doing so for others.
Data about user actions can be logged in real time or collected and sent in batches by the various data providers 108. The individuals 102 visiting the various online sites 150 in turn serve as data sources to data providers 108 and/or for a host 112 of the methods and systems described herein.
Data is conveyed from the data providers 108 to the host 112 periodically. In some cases it is pushed from the data provider 108 to the host 112, while in others it is pulled by the host 112 from the data provider 108. The transfer is typically over a network 110, such as the Internet. The information that gets pulled or pushed across is typically HTTP/HTTPS information for each of the users 102 of that data provider 108, along with demographic information for each of those users 102.
Once the data has come across the network from the data provider 108, the host system 200 looks at individual files. Each data provider 108 follows a convention for the files they send over, which may contain demographic information, clickstream data, and a file signature for all files the data providers send over. To ensure that all of the data was properly transmitted, it is desirable to use a facility such as an MD5 hash of the file. If the size of the file does not match, then the host 112 can determine that the file was not properly transferred.
In embodiments, the host system 200 can comprise a plurality of Free BSD X86 Intel machines, running software written in an object-oriented language such as C++. Referring still to
Once the integrity of the data is confirmed at the step 312, at a step 314 the pre-calculation facility of the host system 200 performs various operations needed to get data from multiple data providers into a common file format, which can be called CLF. Pre-calculation involves various sub-processes disclosed in connection with
After the calculation processes of the step 320 are complete, the system executes various post-calculation processes 322, typically involving merging, sorting and deduplicating results that were obtained by multiple machines into a coherent set of files. Then the system 200 transfers the files at a step 324, such as to a data facility 120, to the client 114, or to a co-location facility. Further details of these processes are disclosed below.
If at the step 404 it is determined that the data provider does not push information, then at a step 420 the system initiates a pull transaction with the data provider. The system then determines at a step 422 whether the files are ready at the data provider. If at the step 422 the files are ready at the data provider, then the system begins the transfer at a step 418. Processing is then sent to a step 424 until transfer is complete. Once transfer is complete (whether the data provider pushes information or the information was pulled by the host), then, at a step 428 the system checks the file signature to confirm that the files were not corrupted during transfer. If the file signature is determined at a step 430 to be okay, then at a step 432 the raw file data is archived. If the file signature is not okay at the step 430, then at the step 414 the operations team and the data provider are alerted that the file needs to be transferred again. Once the data is archived at the step 432, it is determined at a step 434 whether the archiving is complete. If archiving is complete at the step 434, then at a step 438 the files are moved for further processing, such as to an internal network of the host. If there is a problem with the archiving process, then an alert is sent to the operations team at a step 440.
If at the step 422 the files are not ready at the data provider, then at a step 442 the host system waits for a period of time and checks again. A timing facility then determines at a step 444 whether time has expired. If not, then the system continues to wait at the step 442. If so, then at a step 448 an alert is sent to the data provider and the operations team that the desired files were not ready for transfer.
In normal operation the files will transfer properly (in either a push or pull transaction), and the system will arrive at the step 438 with files ready for processing on the internal systems of the host.
Once files are ready for processing, the host system can initiate a pre-calculation process. High level steps of a pre-calculation process are depicted in a flow diagram 500 of
Once the system has determined how many rules apply at the step 804, then, at a step 808, the system applies each applicable rules against the URL. The rules can include filters for various data providers. In embodiments the rules may relate to URLs alone, or may relate to URL/data provider combinations. In embodiments, the rules might relate to the user as well. An example of a rule is as follows. If the data provider is, for example, Amazon.com, it is known that the company modifies its URLs (in a system that is not consistent with some URL protocols) and include session identifiers in the URLs, but not in the query string where they would normally appear for some data providers. By including session identifiers in the URLs, but not in the query string, it appears that each Amazon.com URL is from a unique provider, when in fact all of them are from Amazon.com. To avoid seeing each Amazon URL as coming from a unique provider, the system can apply a rule that pulls out the session identifier from the URL and produces an output URL that doesn't have the session identifier.
Thus, at a step 810, the system outputs a cleansed URL in accordance with the various rules.
Selection of panel members is a statistical process that involves some degree of prediction. For example, selecting a panel involves guessing that a given user will be online again for the next month, so that data can be collected. If the host analyzes data less frequently (e.g., only after the end of a given month), then the host can analyze based on actual information about what users were actually on line in that month. However, if the host wishes to deliver information more rapidly (such as weekly or daily) then it can be preferable to predict what users will be online, rather than waiting to see what users were actually online before making a panel selection.
Next, at a step 924 the system reads the list of panel members generated at the step 922. For a given data file in the CLF format, at a step 928 the panel selection process reads the file line by line and determines whether the user identified by the user id 702 is a panel member. If not, then at a step 930 the system discards the file. If at the step 928 the user id 702 is for a panel member, then at a step 932 the system outputs the file and sends it at a step 934 to the transformation process depicted further in the flow diagram 1000 of
First, at a step 1002, the system sessionizes the data for a given user. Because a user may be logged on to a site all day (or longer), it can be difficult to identify clearly delineated online sessions for the user. In order to sessionize the data, the system looks at data for a particular user for a given day and walks down all of the URLs the user clicked on, looking for inactivity greater than a given time (e.g., 30 minutes) in the timestamps. If there is inactivity greater than the predetermined time, the end of a session is marked. Session identification can be accomplished with conventional log file analysis tools.
Once a session is identified at the step 1002, the system outputs a file in CLF format with a session id added. Thus, the file now has the user ID 702, data provider ID 704, time stamp 708, URL/URLs 710 and protocol 712, as well as the new element, the session ID 1010.
Once the data is divided into session at the step 1002, the system decomposes the URLs in the CLF format file in a step 1004. Decomposition of a URL is accomplished by applying a complex set of rules that are based on the location of given characters at given positions in the URL. Referring to
The URL 1100 is just one of many configurations of URL, and the components 1102 should be understood to be representative of one way of decomposing a URL and labeling constituent elements. Many other ways can be used. In some decompositions, for example, the path 1110, or a combination of the path 1110 and the query string 1112, might just be identified as a query string. In any such embodiment, the decomposition process 1004 breaks down the URL into constituent parts, based on rules that may be adjusted to reflect the URL types of a given data provider or other entity that generates URLs. The decomposition process 1004 can be accomplished in embodiments by a rule engine that steps through the characters of the URL and identifies rules that apply to the given character in the given location, as well as rules that apply to given strings of characters in given components 1102 of the URL. The rule engine essentially figures out the grammar of the URL based on a codification of the rules for all URLs in a wide range of domains.
Maintaining a rule engine that properly decomposes URLs is difficult because of the emergence of facilities such as redirection services that take a given input and redirect the user to a different URL. Thus, the rules must recognize those redirections and classify the redirecting URL as the equivalent of the target URL to which the user is redirected. In addition, many sites (Yahoo.com, for example) are now serving as proxies for other sites, thereby “swallowing” content of the other sites. Thus, if a URL such as the following is visited by a user: http://yahoo.com/proxy/http://ebay.com, the system might initially be tricked into concluding that the domain is yahoo.com. However, the real content for such a URL is actually at another domain, ebay.com. Thus, the URL analysis rules must account for the use of proxies so that they can identify particular page content as being equivalent to other page content, regardless of the use of a proxy in one case and the absence of a proxy in the other. In some situations it may be desirable to identify and store the fact that a page was found through a proxy as well as what the page was. Thus, decomposition could identify and output a file that identifies use of proxies.
The process of decomposing URLs also consists of examining page level content for each of the paths and query strings and identifying rules for what paths and query strings mean in a wide range of contexts.
Once a URL is decomposed at the step 1004 of
The output of the hash step 1008 is a file in a format 1300 that can be called CL5, which is depicted in
Once the hash step 1008 is complete, the transformation process 510 of
In embodiments, the pre-calculation processes of
Once a data file arrives in the system, the system checks file integrity at the step 430 of
Once the pre-calculation process is complete the system can initiate a calculation process. The calculation process is designed to calculate various metrics that the operator of the host system wishes to generate out of the data provided by the data providers. Thus, the calculation process consists of many calculators, one or more for each metric that the operator wishes to calculate. The calculators can operate in parallel fashion running on clusters of machines to facilitate rapid calculation of the various metrics. Thus, in the calculation process the system ends up with multiple intermediate files running across multiple servers.
Once the calculation processes are complete, another process can be kicked off. That post-calculation process picks up the multiple files across multiple servers to merge, sort, and deduplicate files. Essentially, the post-calculation process merges data from the various servers that performed the calculation processes.
After post-calculation processes are complete, the resulting files can be sent out to various locations, such as a co-location server, where the files may be accessed by clients of the host system, and/or to an internal data warehouse, to allow analysts of the host to look at them. Depending on how the host decides to partition a day's data there might be multiple data warehouses where files will reside. Once there, users can request data from the warehouse, which they do through a facility such as a broker (a web-based interface that takes a request, decomposes it, identifies where in the data warehouse group of machines the data resides, and returns the data to the user through the broker). In embodiments the broker may give an intermediate result or may provide a final result that is the result of the sorting, merging and deduplicating processes. In embodiments the broker can allocate what files go on what machines and then return the files when requested by users.
More details of the calculation process are provided below. The calculation process may take place periodically, whether the period is daily, weekly, monthly or on some other time schedule. In embodiments, calculations are done daily to provide clients of the host with timely information. The calculation process can calculate many different metrics, such as the number of unique visitors to a page of content, search terms used by users, and classification of page content by industry verticals.
Next, at a step 1404, a checksum process happens for the daily (or other period) CL5 files to tell whether the pre-calculation process is finished writing the files out or not. When files are completed, they can be appended with a suffix, e.g., Filename.done. In the files are MD5 checksums of the files, so the system looks for the .done suffix and then compares the MD5 signature to the checksum to determine whether the files are ready to process. If the checksum works, then the system indicates so by a CHECKSUM OK message.
Next, at a step 1408, a job or batch handling facility manages a set of processing nodes to handle processing of the files. The job/batch handling facility is programmed to know which nodes are free (available for processing), which ones are busy, and which calculations can run on which nodes. (Some calculations can only run on certain nodes, while some can run on any node).
Once the handling facility hands of the processing to the nodes, the system checks periodically to determine whether the processing of a given metric is completed. For example, if the metric is the number of unique visitors to a URL, the system can at a step 1410 check to see whether the calculation of “uniques” is completed. If calculation for a metric is not complete, then processing continues. When the metric is complete at the step 1410, the system initiates processing for the next metric at a step 1412. Of course the system may calculate different metrics simultaneously, rather than completing one metric and then completing another. The flow diagram 1400 should be understood to set out the high level steps for only one embodiment of batch processing and identifying completion of a given metric calculation. Other embodiments would be readily understood by one of ordinary skill in the art.
In embodiments, each day the job processing facility knows how many work orders it has in the system. It looks up the work orders for that day. When it finds files ready to run, it begins to hand them to the nodes to run. The work orders can be conditional. Thus, they can run a given metric if the checksum is ok for the incoming file, and they can process a given metric conditioned on completion of another metric. For a work order, the job processing facility is programmed to know how many machines it needs, and it reserves the machines it needs. If processing a given metric, such as “uniques” requires four nodes, then the job processing facility can start running the calculation of uniques on four nodes, putting the other nodes on “wait” status. The job processing facility does this for each metric. The job processing facility can be programmed with a language of its own, similar to a compiler.
At a high level, the calculation process for a given metric checks to see if files are in the proper location, reserves machines needed for that metric, initiates processes on the appropriate machines, and outputs a file. Referring to
Once the job/batch processing facility completes calculations for a given metric, a rollup process of the host system 200 can determine how many files were produced for that metric that day (or period), then combine them. For most metrics that can be accomplished by simply merging files. For some metrics, such as unique visitors, simply merging files may not be enough, because there may be a need to de-duplicate the files to avoid double counting a given user's visit to a URL.
Once the rollup process is complete the machine can produce a file (e.g., “date.unique”) that contains the data for that metric for that date. In embodiments the file can be a GZIP file.
In an embodiment, the external data is data from a dictionary of search terms from a multiplicity of web sites. In an embodiment, the host collects search terms from more than 27,000 sites for use by the search terms metric calculator. To know what the search pages and types are, the operator of the host system 200 or a third party goes to various sites and to sections within sites and figures out what the search URL looks like for that site. For example, for Yahoo.com the search URL might look like: search.yahoo.com? There are a wide variety of formulations for search strings. Some take the basic form of “domain.com/p=query string,” but others take different forms. For example, some sites are co-branded by two different parties. For example, if google.com and the Washington Post co-brand a site, the URL may look as follows: “Wp.google.com (parameter)p=washingtonpost.com.” The dictionary of search terms must recognize this as a Washington post search string, not a google.com one. Thus, the system sets up a rule that recognizes that if the string is of the form xx.google.com/xxx? q=xxx, the credit goes to the domain identified by the query string, not to google.com.
By way of example, Yahoo.com has hundreds of search pages on their site (finance ticker, groups, mail, etc.), so the host operator can establish or access a dictionary that takes the various forms and translates them to obtain the semantic meaning, or the host can use a rule or similar facility to accomplish a similar function.
So, when the job runs at the node level, the search terms calculator reads the external data, doing lookups on tables (held in node machines in memory) to figure out who should get credit for what URL. As with rules used in pre-calculation, it is too difficult to compare each URL with all possible search terms in the dictionary. Instead, the system can pass each line a single time, and the calculators figure out whether they should calculate something with the line or throw it out.
As with the other metrics, at a step 1710 the system rolls up the files into a single file ready for transfer to the data facility 120, the client 114 or the co-location facility 118.
Next, at a step 1808, the system accesses an external communications facility. Thus, the machines used for the verticals calculation must be able to access an external communications facility, which will be used to reach an external data set. This time the system must go outside the internal network and do something in order to allow the processing.
The term “verticals” refers to industry classifications. For example, a “vertical” might be business-to-business services, with subcategories ranging from financial services, to office services, etc. Another set of verticals might be business-to-consumer services, with verticals including retail, financial services, etc. The system can classify any number of verticals.
Once the page extractor process 1810 is complete, the fetcher module 1908 is engaged to run a fetcher process in a step 1812. The fetcher process reads a line in the file for a given site and determines whether it needs to update the data for that site. If so, then the fetcher process 1812 downloads the identified page. If the process has seen the identified page before, it can skip to the next one. If the fetcher process finds a new page, it downloads the page and puts it in the director of pages for that site. In embodiments the fetcher process only runs on certain nodes, because the host system must avoid flooding a site with downloads. The fetcher accesses an external network and in embodiments only runs on one or a few machines. The fetcher module can be programmed to now the maximum rate at which it should pull pages. In embodiments the fetcher module knows how to take codes for changed pages, moved pages, and indications that a site is not in operation. In each case it can produce alerts for the operator of the host system 200 to change something.
The next step in the verticals metrics calculation engages the classifier module 1910 is a classifying step 1814. Once the fetcher module 1908 has downloaded any new page identifiers, then for every URL in a given file (e.g., a file for Office Depot pages), the classifier module 1910 writes out a record to a fetcher database. In embodiments this can be a flat file hierarchy held locally and mirroring the live site, e.g.—officedepot.com/category/product/ID. Such a file can serve as an archive of the hierarchy of pages at the given site, including, for example, product directory files for all products offered at the site. The classifier module 1910 figures out for that day how many unique products it saw. In embodiments, it goes into flat file database and knows how to open the identified page and pull out the product information (title, description, price, shipping status, etc.). How to do that for a given site can be stored in the configuration file for the page extractor. It may be different for every site and page that the host system 200 tracks. Thus, the host system operator may have to come up with different patterns for the various pages and sites that it covers.
The classifying step 1814 takes the information and writes out another file. An example of a file structure would include elements such as the product name, the price, the quantity obtained for that price, the user ID, and the date. For example, it might appear as follows: Red stapler/$10.00/quantity 2/ID/date. The classifier module 1910 can do this for all of the different pages that are visited by users in the CL5 files for all of the CL5 files. In embodiments the classifier module 1910 can aggregate records across a category (e.g., B2B: Office). In such embodiments, a user of the data can compare products, such as all staplers, to see which ones users are viewing and buying. The classifying step 1814 can do this for office sites, retail, financial services, stock tickers, and other pages. In embodiments it can map a ticker back to the real name for the company.
Once a classifying step 1814 is complete, then at a step 1818 the system can rollup the files for the verticals calculation, as it does for the other metrics that are involved in job/batch processing.
The methods and systems described above cover three out of many possible metrics that can be calculated. Some metrics may be run with data just from the CL5 files that are generated in the pre-calculation processes disclosed above. Others require reference to an external data set as well as data from the CL5 files. Still others require access to page information, such as fetched from the Internet with the fetcher module 1908. Examples of metrics include, without limitation, unique visitors, search terms, verticals, cross-correlations between sites, uses of particular patterns of pages, and many others. Each different metric can have a unique calculator, typically running on a node of the cluster 214, using a program coded for that metric.
The various metrics can be used by analysts to deduce information that is relevant to many different kinds of business processes. For example, data about how users interact with pages that relate to applications for a quote on an insurance policy can produce insight into what is involved in converting an insurance shopper into an insurance buyer.
The data obtained can be quite comprehensive. In embodiments, it may consist of user interactions with many thousands of pages divided into hundreds or thousands of vertical categories.
The system 200 may further have a post-calculation facility 212 for merging, sorting and deduplicating the files that are generated in the calculation process. After the post-calculation facility 212 completes these tasks, the files can be deployed internally to data warehouses, or externally to external data facilities, such as a co-location facility 118 that can be accessed by a client.
Data that is collected and processed by the methods and systems disclosed herein can be used as a basis for modifying a wide range of business processes. One area is in the promotion of goods and services. It is well known that demographic information can be used as a basis for designing product promotion; that is, certain groups of customers are more likely to purchase certain goods and services than others. For example, parents are more likely than teenagers to purchase baby strollers. Other information about user behavior can be used to assist in designing product promotion as well.
One embodiment of the method and systems disclosed herein is an improved method of promoting automobiles. Automobiles are typically promoted with incentives, such as low-interest financing, rebates, and the like. Presenting the right promotion at the right time increases the likelihood of making a sale while minimizing the cost of promotions. Understanding what vehicles a particular car model is being cross-shopped against regionally enables the design of incentives to take account of the varying competitive set. For example, if a customer shops for a Toyota Camry, is that customer more likely to look at a Nissan Maxima, or a Nissan Altima, either of which could be viewed as similar. Knowing what cross-shopping is likely to occur allows a vendor to present incentives in a manner that accounts for the likely comparisons that the customer will make between the vendor's products and services and potentially competitive ones. The methods and systems disclosed herein allow an analyst to track cross-shopping behavior and allow a vendor to modify incentives in response to inferences drawn from cross-shopping behavior. By identifying regional variations and changes in the vehicles cross-shopped, an analyst can recommend incentives to fit the changed circumstances, thereby maximizing sales at the lowest incentive cost.
A second example would be brand positioning. One relevant factor in presenting a marketing message is the order in which the user has shopped various product offerings. For example, if an offering is the first one that a user has shopped, then it should be advertised in a fashion that encourages users to purchase immediately, rather than consider other offers. If the user has already seen more than one offer, then the subsequent offers need to be positioned differently. If a vendor is presenting an offer after many others have been considered, then the offer needs to be sufficiently arresting to get the attention of customer late in the process. So, for purposes of branding, messaging, and marketing communications, it is valuable for a vendor to know whether that vendor is typically the first vendor shopped, or whether the vendor typically comes later in the process. This is true whether the marketing message is presented online or offline.
By identifying patterns of online shopping for the products or services offered by a particular vendor, the methods and systems disclosed herein allow an analyst to determine the order in which a customer typically shops for products or services of that type. Then the vendor can select and present messages, promotions, and branding that is consistent with the position of the vendor in the typical shopping pattern. This approach can be used in any product or service sector, ranging from automotive, to retail, financial, and others.
Knowing customer patterns can also be useful for other business purposes, such as demand forecasting. For example, analysis of customer interest can determine if a predictable share of customers who have looked at a set of competitive products and services will purchase a particular vendor's product or service. Changes in the share of online users who research goods or services can be used to predict change in market share offline. Thus, by allowing tracking changes in online market share, the methods and systems disclosed herein can assist in offline business planning for changes that depend on changes in sales or market share, such as manufacturing and inventory needs.
Thus, data about online behavior can also help in forecasting demand. For example, in the auto market, knowing what people are researching (e.g., measuring frequency, recency and duration of online shopping for particular models), allows the business to forecast demand for particular make, model and trim in particular geographic regions. That forecast can be used allow a business to set inventory and manufacturing at optimal levels. Also online interest share changes (showing interest in a particular product) can be correlated to offline sales. Then the business can manage its supply chain and manufacturing to allocate the correct amounts of inventory to the right places at the right time.
Another area in which tracking online user behavior can allow strategic modification of offline business processes is in the area of retail store operation. Retail stores are designed to encourage users to see merchandise in a particular pattern. Stores are particularly designed to present complementary products near each other. For example, tortilla chips and salsa are usually presented in the same area of a supermarket. Also, the order in which a customer sees merchandise can make a difference in the likelihood the customer will buy a particular item. For example, studies have shown that people perceive the price of an item to be lower if they have previously been presented with a higher-priced item. For example, a forty-dollar tie seems less expensive if the customer has just selected a five hundred dollar suit. Moreover, the need for the tie may be more clear to the customer if the decision to purchase the suit has already occurred.
It is expensive and time consuming to rearrange a retail store to try to figure out which items should be presented together, and what order should be used to present items. In embodiments of the methods and systems disclosed herein, tracking user behavior online allows an analyst to draw an inference about what items should be presented together, and what items should be presented in what order, in a retail store. For example, if it is discovered that most online users who purchase personal computers also shop for printers, then an analyst might conclude that computers and printers should be presented together. In some cases the connection between the items might have been obvious, but in other situations a correlation between shopping for one item and shopping for another may not be readily apparent without the ability to track and recognize statistical correlations among items, a process that is made possible by tracking user behavior in the methods and systems disclosed herein. Thus, the methods and systems disclosed herein allow the user to organize merchandise in the physical world based on online behavior.
In addition to locating merchandise within a retail environment, it is also possible to draw inferences about how to locate an entire store based on analysis of user behavior online. For example, if goods and services of two different vendors are identified through statistical techniques as complementary, then one vendor may wish to select a location near the other vendor, even if the connection between the two is not intuitively obvious.
In other embodiments, data from online behavior can be used to draw inferences to develop offline business processes in the financial services area. For example, businesses need to determine what products and services should be offered through what distribution channel, whether it is an online channel, a call center, a branch, or an ATM channel. For example, financial advice and stock quotes might best be delivered through different channels. If online behavior shows that customers examine individual stock information online, but get confused with mutual funds, then a business can decide to put the individual stock information online, but to offer mutual funds through other channels, such as advisors in branch offices. Thus, figuring out the proper offline distribution channel can be done with the assistance of inferences drawn from online behavior.
Location decisions also benefit from online information. For example, if a business can learn where its competitor's online shoppers come from by zip code, a business can decide to locate its physical facilities in locations where the competitor's online shopper density is highest or lowest, as desired. This can help the business get new business from competitors, rather than cannibalizing its own customer base.
Other retail business processes can also be helped. By determining interest share by zip code, a catalog provider can, for example, drop the customized versions of a catalog by zip code to meet regional needs. Similarly, other direct marketing campaigns, such as mail campaigns, can be targeted to the right demographic and location based on online interest.
Pricing decisions also benefit from analysis of online behavior. For example, knowing how the number of transactions is affected when a price is changed can help a business determine an optimal price for goods or services. It can be also useful to know how many people shop elsewhere after a price increase, as compared to before. By looking at behavior across multiple sites it is possible to set prices at the highest point of price indifference. Thus prices can be tuned for online and offline offerings based on online behavior data.
It is also possible to watch behavior of a competitor's customers after the competitor increases prices. Also, a business can introduce price changes online to a limited set of customers to determine sensitivity before making more extensive price changes offline.
The methods and systems disclosed herein can assist media businesses as well. By determining the patterns and reasons for online migration between different online content sources, it is possible to draw inferences about the customer's behavior that allow the content to be designed to keep the user at a given content source. For example, if a readers of Forbes regularly depart its online site to look at particular portions of Fortune's online site, it is possible to redesign the content of the Forbes site to contain similar content, thus diminishing the chance the reader will depart. Thus, behavior online can be used to help redesign media content to retain customers. Media businesses can also use online behavior to identify geographic areas of interest and other demographic areas of interest for books, music, news, film and other content. In turn, offline channels of distribution for media content can be altered to target particular geographic or demographic groups. Also, media businesses can target attracting customers based on their online behavior profile with a view to maximizing the attraction of the “best” or most loyal customers.
It is also possible to forecast demand for media content. For example, web traffic at particular sites related to upcoming movie content may tell film distributors which films should be shown on which screens and for how long. Media providers can also determine what kind of related merchandise to sell along with the media, and what kinds of promotions to engage in. The data can be used to determine appropriate franchising arrangements as well, such as how much to charge for a franchise to distribute toys or other novelties that reflect characters in a film.
Similar forecasting can be used before release of a new musical composition, book, or other media work.
All patents, patent applications, technical specifications, books, and other documents identified herein are hereby incorporated by reference. While the invention has been described in connection with certain preferred embodiments, other embodiments would be readily understood by one of ordinary skill in the art and are encompassed herein.
|Patente citada||Fecha de presentación||Fecha de publicación||Solicitante||Título|
|US5583763||9 Sep 1993||10 Dic 1996||Mni Interactive||Method and apparatus for recommending selections based on preferences in a multi-user system|
|US5749081||6 Abr 1995||5 May 1998||Firefly Network, Inc.||System and method for recommending items to a user|
|US5893077||23 Ago 1995||6 Abr 1999||Microsoft Corporation||Method and apparatus for generating and collecting a billing event object within an on-line network|
|US5930764||23 Ago 1996||27 Jul 1999||Citibank, N.A.||Sales and marketing support system using a customer information database|
|US6055570||3 Abr 1997||25 Abr 2000||Sun Microsystems, Inc.||Subscribed update monitors|
|US6216135||17 Ago 1999||10 Abr 2001||Siebel Systems, Inc.||Method of determining visibility to a remote database client of a plurality of database transactions having variable visibility strengths|
|US6314451||13 Jul 1999||6 Nov 2001||Unicast Communications Corporation||Ad controller for use in implementing user-transparent network-distributed advertising and for interstitially displaying an advertisement so distributed|
|US6334108||21 Oct 1996||25 Dic 2001||Catalina Marketing International, Inc.||Method and system for selective incentive point-of-sale marketing in response to customer shopping histories|
|US6334110||10 Mar 1999||25 Dic 2001||Ncr Corporation||System and method for analyzing customer transactions and interactions|
|US6448981||9 Dic 1997||10 Sep 2002||International Business Machines Corporation||Intermediate user-interface definition method and system|
|US6519571||27 May 1999||11 Feb 2003||Accenture Llp||Dynamic customer profile management|
|US6574587||27 Feb 1998||3 Jun 2003||Mci Communications Corporation||System and method for extracting and forecasting computing resource data such as CPU consumption using autoregressive methodology|
|US6606304||21 May 1999||12 Ago 2003||On Guard Plus||System for real-time monitor and response|
|US6721749||6 Jul 2000||13 Abr 2004||Microsoft Corporation||Populating a data warehouse using a pipeline approach|
|US6820116||21 Dic 2001||16 Nov 2004||Nokia Corporation||Mobile browsing booster system|
|US6834372||10 Feb 2000||21 Dic 2004||International Business Machines Corporation||Internet web browser with proximity sensitie hyperlink history report|
|US6839680||30 Sep 1999||4 Ene 2005||Fujitsu Limited||Internet profiling|
|US6847969||3 May 2000||25 Ene 2005||Streetspace, Inc.||Method and system for providing personalized online services and advertisements in public spaces|
|US6904408||19 Oct 2000||7 Jun 2005||Mccarthy John||Bionet method, system and personalized web content manager responsive to browser viewers' psychological preferences, behavioral responses and physiological stress indicators|
|US6934687||20 Nov 1997||23 Ago 2005||Ncr Corporation||Computer architecture and method for supporting and analyzing electronic commerce over the world wide web for commerce service providers and/or internet service providers|
|US6934748||25 Ago 2000||23 Ago 2005||Memetrics Holdings Pty Limited||Automated on-line experimentation to measure users behavior to treatment for a set of content elements|
|US6981040||20 Jun 2000||27 Dic 2005||Utopy, Inc.||Automatic, personalized online information and product services|
|US7032168||27 Nov 2001||18 Abr 2006||Be Free, Inc.||Method and apparatus for generating website links|
|US7035855||6 Jul 2000||25 Abr 2006||Experian Marketing Solutions, Inc.||Process and system for integrating information from disparate databases for purposes of predicting consumer behavior|
|US7243075||3 Oct 2000||10 Jul 2007||Shaffer James D||Real-time process for defining, processing and delivering a highly customized contact list over a network|
|US7249032||30 Nov 2001||24 Jul 2007||Demandtec Inc.||Selective merchandise price optimization mechanism|
|US7272629||3 Jul 2002||18 Sep 2007||Yamaha Corporation||Portal server and information supply method for supplying music content of multiple versions|
|US7386466||17 Jun 2002||10 Jun 2008||Tvc International Inc.||Continuously updated data processing system and method for measuring and reporting on value creation performance that supports real-time benchmarking|
|US7478035||31 Oct 2000||13 Ene 2009||Eclarity, Inc.||Verbal classification system for the efficient sending and receiving of information|
|US20010014868||22 Jul 1998||16 Ago 2001||Frederick Herz||System for the automatic determination of customized prices and promotions|
|US20010032115||22 Dic 2000||18 Oct 2001||Michael Goldstein||System and methods for internet commerce and communication based on customer interaction and preferences|
|US20010051932||13 Mar 2001||13 Dic 2001||Kannan Srinivasan||Method and system for dynamic pricing|
|US20020004868||5 Jul 2001||10 Ene 2002||Nec Corporation||Parallel processing system in which use efficiency of CPU is improved and parallel processing method for the same|
|US20020010757||1 Dic 2000||24 Ene 2002||Joel Granik||Method and apparatus for replacement of on-line advertisements|
|US20020016731||25 May 2001||7 Feb 2002||Benjamin Kupersmit||Method and system for internet sampling|
|US20020035498||30 Mar 2001||21 Mar 2002||Kehoe Jeffrey W.||Filtering web proxy for recording web-based transactions that supports secure HTTP steps|
|US20020038350||30 Abr 2001||28 Mar 2002||Inceptor, Inc.||Method & system for enhanced web page delivery|
|US20020042733||22 Feb 2001||11 Abr 2002||Lesandrini Jay William||Enhancements to business research over internet|
|US20020042821||10 May 2001||11 Abr 2002||Quantified Systems, Inc.||System and method for monitoring and analyzing internet traffic|
|US20020052782||26 Jun 2001||2 May 2002||Mark Landesmann||Buyer-driven targeting of purchasing entities|
|US20020053078||18 Abr 2001||2 May 2002||Alex Holtz||Method, system and computer program product for producing and distributing enhanced media downstreams|
|US20020065912||30 Nov 2001||30 May 2002||Catchpole Lawrence W.||Web session collaboration|
|US20020095387||13 Mar 2002||18 Jul 2002||Bertrand Sosa||Online content portal system|
|US20020099824||24 Oct 2001||25 Jul 2002||Bender Brad H.||Method and system for sharing anonymous user information|
|US20020100042||29 Mar 2002||25 Jul 2002||Denis Khoo||Method and system for providing intelligent advertisement placement in a motion picture|
|US20020103680||31 Ene 2002||1 Ago 2002||Newman Les A.||Systems, methods and computer program products for managing employee benefits|
|US20020107730||5 Feb 2001||8 Ago 2002||Robert Bernstein||Method and apparatus for identifying customers for delivery of promotional materials|
|US20020116348||18 May 2001||22 Ago 2002||Phillips Robert L.||Dynamic pricing system|
|US20020116531||21 Feb 2001||22 Ago 2002||International Business Machines Corporation||Applying anonymous personalization to web-based customer interactions|
|US20020123359||30 Nov 2001||5 Sep 2002||Multiscience System Pte Limited||Network for information transfer for mobile stations|
|US20020123926||1 Mar 2001||5 Sep 2002||Bushold Thomas R.||System and method for implementing a loyalty program incorporating on-line and off-line transactions|
|US20020196273||24 Jun 2002||26 Dic 2002||Krause Philip R.||Method and apparatus for consolidating network information|
|US20020198791||28 Ene 2002||26 Dic 2002||Perkowski Thomas J.||Internet-based consumer product brand marketing communication system which enables manufacturers, retailers and their respective agents, and consumers to carry out product-related functions along the demand side of the retail chain in an integrated manner|
|US20020198882||15 Ene 2002||26 Dic 2002||Linden Gregory D.||Content personalization based on actions performed during a current browsing session|
|US20030018677||6 Mar 2002||23 Ene 2003||Ashish Mathur||Increasing precision in multi-stage processing of digital signals|
|US20030023715||16 Jul 2001||30 Ene 2003||David Reiner||System and method for logical view analysis and visualization of user behavior in a distributed computer network|
|US20030037124||3 Jul 2002||20 Feb 2003||Atsushi Yamaura||Portal server and information supply method for supplying music content|
|US20030050815||11 Abr 2002||13 Mar 2003||Seigel Ronald E.||System for purchasing geographically distinctive items via a communications network|
|US20030110072||4 May 2001||12 Jun 2003||Demand Tec. Inc.||Interface for merchandise promotion optimization|
|US20030126095||28 Jun 2002||3 Jul 2003||Docomo Communications Laboratories Usa, Inc.||Context-aware market-making service|
|US20030126250||13 Dic 2000||3 Jul 2003||Neeraj Jhanji||Systems for communicating current and future activity information among mobile internet users and methods therefor|
|US20030128818||19 Sep 2002||10 Jul 2003||Telemessenger Limited||Information transmission system|
|US20030135405||17 Ene 2002||17 Jul 2003||Vehnet, Inc.||System, method, application to maximize electronic commerce and sales in retail automotive industry|
|US20030149696||7 Feb 2002||7 Ago 2003||Steve Nelson||Internet based system for creating presentations|
|US20030154162||11 Feb 2003||14 Ago 2003||Danaher John Thomas||Credit report retrieval system including voice-based interface|
|US20040064449||14 Jul 2003||1 Abr 2004||Ripley John R.||Remote scoring and aggregating similarity search engine for use with relational databases|
|US20040107363||22 Ago 2003||3 Jun 2004||Emergency 24, Inc.||System and method for anticipating the trustworthiness of an internet site|
|US20040205119||26 Mar 2002||14 Oct 2004||Streble Mary C.||Method and apparatus for capturing web page content development data|
|US20040249938||17 Jun 2004||9 Dic 2004||Bunch Clinton D.||System and method for monitoring access to a network by a computer|
|US20050222906||8 Feb 2005||6 Oct 2005||Chen Timothy T||System and method of targeted marketing|
|US20050234922||2 Mar 2005||20 Oct 2005||Parekh Sanjay M||Systems and methods for determining, collecting, and using geographic locations of internet users|
|US20050235318||16 Jun 2005||20 Oct 2005||Grauch Edward R||Method and system for tracking network use|
|US20060248452||30 Jun 2006||2 Nov 2006||Inceptor, Inc.||Method and system for enhanced Web page delivery and visitor tracking|
|US20060253434||10 Jul 2006||9 Nov 2006||Beriker James K||Auction-based search engine|
|US20070043817||27 Oct 2006||22 Feb 2007||MailFrontier, Inc. a wholly owned subsidiary of||Personalized electronic-mail delivery|
|US20070061313||3 Nov 2006||15 Mar 2007||Brewster Kahle||Detection of search behavior based associations between web sites|
|US20070198486||24 Ago 2006||23 Ago 2007||Daniel Abrams||Internet search engine with browser tools|
|US20080071767||21 Mar 2007||20 Mar 2008||Semdirector, Inc.||System and method for measuring the effectiveness of an on-line advertisement campaign|
|US20080177778||24 Oct 2007||24 Jul 2008||David Cancel||Presentation of media segments|
|US20080177779||24 Oct 2007||24 Jul 2008||David Cancel||Presentation of media segments|
|US20080183664||24 Oct 2007||31 Jul 2008||David Cancel||Presenting web site analytics associated with search results|
|US20080183718||24 Oct 2007||31 Jul 2008||Man Jit Singh||Clickstream analysis methods and systems|
|US20080183796||24 Oct 2007||31 Jul 2008||Man Jit Singh||Clickstream analysis methods and systems|
|US20080183805||24 Oct 2007||31 Jul 2008||David Cancel||Presentation of media segments|
|US20080183806||24 Oct 2007||31 Jul 2008||David Cancel||Presentation of media segments|
|US20080183867||24 Oct 2007||31 Jul 2008||Man Jit Singh||Clickstream analysis methods and systems|
|US20080183868||24 Oct 2007||31 Jul 2008||Man Jit Singh||Clickstream analysis methods and systems|
|US20080183869||24 Oct 2007||31 Jul 2008||Man Jit Singh||Clickstream analysis methods and systems|
|US20080183870||24 Oct 2007||31 Jul 2008||Man Jit Singh||Clickstream analysis methods and systems|
|US20080189253||29 Oct 2007||7 Ago 2008||Jonathan James Oliver||System And Method for Adaptive Text Recommendation|
|US20080189254||12 Nov 2007||7 Ago 2008||David Cancel||Presenting web site analytics|
|US20080189408||12 Nov 2007||7 Ago 2008||David Cancel||Presenting web site analytics|
|US20080250026||3 Mar 2008||9 Oct 2008||Linden Gregory D||Recommendations based on cross-site browsing activities of users|
|US20080288635||4 May 2008||20 Nov 2008||Jaye Daniel J||System and method for associating a client identity between servers|
|US20080301281||31 May 2007||4 Dic 2008||Microsoft Corporation||Search Ranger System and Double-Funnel Model for Search Spam Analyses and Browser Protection|
|US20100030894||30 Jun 2009||4 Feb 2010||David Cancel||Computer program product and method for estimating internet traffic|
|1||"U.S. Appl. No. 11/923,454, Final Office Action mailed Aug. 19, 2009", , 12 Pgs.|
|2||"U.S. Appl. No. 11/923,454, Non-Final Office Action mailed Apr. 5, 2010", , 13.|
|3||"U.S. Appl. No. 11/923,454, Non-Final Office Action mailed Jan. 8, 2009", , 7 pgs.|
|4||"U.S. Appl. No. 11/923,470, Non-Final Office Action mailed Aug. 28, 2009", , 7 pgs.|
|5||"U.S. Appl. No. 11/923,470, Non-Final Office Action mailed Feb. 22, 2010", , 13.|
|6||"U.S. Appl. No. 11/923,485, Non-Final Office Action mailed Aug. 13, 2009", , 12 pgs.|
|7||"U.S. Appl. No. 11/923,486, Final Office Action mailed Mar. 3, 2010", , 12.|
|8||"U.S. Appl. No. 11/923,506, Final Office Action mailed Feb. 26, 2010", , 21.|
|9||"U.S. Appl. No. 11/923,506, Non-Final Office Action mailed Aug. 25, 2009", , 15 Pgs.|
|10||"U.S. Appl. No. 11/923,560, Non Final Office Action mailed Oct. 28, 2009", , 9.|
|11||"U.S. Appl. No. 11/923,609 Non-Final Office Action mailed Jan. 15, 2009", , 7 Pgs.|
|12||"U.S. Appl. No. 11/923,609, Final Office Action mailed Aug. 7, 2009", , 11 Pgs.|
|13||"U.S. Appl. No. 11/923,609, Notice of Allowance mailed Jan. 29, 2010", , 14 Pgs.|
|14||"U.S. Appl. No. 11/923,610, Non Final Office Action mailed Sep. 2, 2009", , 19.|
|15||"U.S. Appl. No. 11/923,610, Notice of Allowance mailed Apr. 29, 2010", , 7.|
|16||"U.S. Appl. No. 11/923,617, Non-Final Office Action mailed Sep. 1, 2009", , 11 Pgs.|
|17||"U.S. Appl. No. 11/923,617, Notice of Allowance mailed Mar. 23, 2010", , 7.|
|18||"U.S. Appl. No. 11/923,620, Final Office Action mailed Apr. 23, 2010", , 29 PGS.|
|19||"U.S. Appl. No. 11/923,620, Non-Final Office Action mailed Sep. 29, 2009", , 19 Pgs.|
|20||"U.S. Appl. No. 11/923,621, Final Office Action mailed Apr. 23, 2010", , 29 Pages.|
|21||"U.S. Appl. No. 11/923,621, Non-Final Office Action mailed Sep. 30, 2009", , 19 Pgs.|
|22||"U.S. Appl. No. 11/923,624, Final Office Action mailed Apr. 23, 2010", , 28.|
|23||"U.S. Appl. No. 11/923,624, Non-Final Office Action mailed Sep. 29, 2009", , 19 Pgs.|
|24||"U.S. Appl. No. 11/938,710, Notice of Allowance mailed Feb. 17, 2010", , 8 Pgs.|
|25||"U.S. Appl. No. 11/938,716, Non-Final Office Action mailed Mar. 12, 2009", , 6 pgs.|
|Patente citante||Fecha de presentación||Fecha de publicación||Solicitante||Título|
|US8135833||1 Jun 2011||13 Mar 2012||Compete, Inc.||Computer program product and method for estimating internet traffic|
|US8171156 *||24 Jul 2009||1 May 2012||JumpTime, Inc.||Method and system for determining overall content values for content elements in a web network and for optimizing internet traffic flow through the web network|
|US8356097||27 Ene 2012||15 Ene 2013||Compete, Inc.||Computer program product and method for estimating internet traffic|
|US8626834||27 Sep 2010||7 Ene 2014||Compete, Inc.||Clickstream analysis methods and systems related to modifying an offline promotion for a consumer good|
|US8719273 *||26 Ago 2011||6 May 2014||Adobe Systems Incorporated||Analytics data indexing system and methods|
|US8768943 *||1 Sep 2010||1 Jul 2014||International Business Machines Corporation||Multi-source consumer behavior tracking system|
|US8954580||14 Ene 2013||10 Feb 2015||Compete, Inc.||Hybrid internet traffic measurement using site-centric and panel data|
|US9037637||6 Feb 2012||19 May 2015||J.D. Power And Associates||Dual blind method and system for attributing activity to a user|
|US9123056||26 Nov 2013||1 Sep 2015||Compete, Inc.||Clickstream analysis methods and systems related to modifying an offline promotion for a consumer good|
|US9177326||21 Feb 2012||3 Nov 2015||OpenX Technologies, Inc.||Method and system for determining overall content values for content elements in a web network and for optimizing internet traffic flow through the web network|
|US9292860||7 Jul 2015||22 Mar 2016||Compete, Inc.||Clickstream analysis methods and systems related to modifying an offline promotion for a consumer good|
|US9501781||14 Mar 2016||22 Nov 2016||Comscore, Inc.||Clickstream analysis methods and systems related to improvements in online stores and media content|
|US9508082 *||3 Oct 2012||29 Nov 2016||Groupon, Inc.||Offline location-based consumer metrics using online signals|
|US20100146144 *||24 Jul 2009||10 Jun 2010||Jump-Time, Inc.||Method and System for Determining Overall Content Values for Content Elements in a Web Network and for Optimizing Internet Traffic Flow Through the Web Network|
|US20120054213 *||1 Sep 2010||1 Mar 2012||Hemanth Puttaswamy||Multi-source consumer behavior tracking system|
|US20170140419 *||24 Oct 2016||18 May 2017||Groupon, Inc.||Offline location-based consumer metrics using online signals|
|Clasificación de EE.UU.||709/201, 705/16, 709/225, 709/219, 709/224, 726/17, 726/2, 709/226, 726/27, 705/26.8|
|Clasificación internacional||G06Q30/00, G06F15/16|
|Clasificación cooperativa||H04L67/02, H04H60/46, G06Q30/0204, G06F17/30861, G06F15/17, G06F15/16, G06Q30/0244, H04H60/45, G06Q20/201, G06Q20/203, G06Q20/20, G06Q30/0633, G06Q30/0251, G06Q30/0241, G06Q30/02, G06Q20/206, G06Q30/0224|
|Clasificación europea||G06Q30/02, G06Q30/0633, G06Q20/206, G06Q20/203, G06Q20/20, G06Q20/201, G06Q30/0241, G06Q30/0251, G06Q30/0224|
|6 Oct 2009||AS||Assignment|
Owner name: COMPETE, INC., MASSACHUSETTS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:SINGH, MAN JIT;CANCEL, DAVID;REEL/FRAME:023331/0858;SIGNING DATES FROM 20030123 TO 20030127
Owner name: COMPETE, INC.,MASSACHUSETTS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:SINGH, MAN JIT;CANCEL, DAVID;SIGNING DATES FROM 20030123TO 20030127;REEL/FRAME:023331/0858
Owner name: COMPETE, INC., MASSACHUSETTS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:SINGH, MAN JIT;CANCEL, DAVID;SIGNING DATES FROM 20030123TO 20030127;REEL/FRAME:023331/0858
|1 Feb 2011||CC||Certificate of correction|
|4 Sep 2012||AS||Assignment|
Owner name: COMPETE, INC., MASSACHUSETTS
Free format text: ADDRESS CHANGE;ASSIGNOR:COMPETE, INC.;REEL/FRAME:028896/0934
Effective date: 20120820
|10 Sep 2012||AS||Assignment|
Owner name: COMPETE, INC., MASSACHUSETTS
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ASSIGNEE ADDRESS PREVIOUSLY RECORDED ON REEL 028896 FRAME 0934. ASSIGNOR(S) HEREBY CONFIRMS THE ASSIGNMENT;ASSIGNOR:COMPETE, INC.;REEL/FRAME:028925/0516
Effective date: 20120820
|10 Feb 2014||FPAY||Fee payment|
Year of fee payment: 4
|6 May 2016||AS||Assignment|
Owner name: COMSCORE, INC., VIRGINIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:COMPETE, INC.;REEL/FRAME:038630/0725
Effective date: 20160428